The study employs a mini-computer based medical record system developed inhouse by the investigators. This computerized medical record system can accept and interpret English-like statements of idealized medical logic dealing with a subset of the medical problems faced by the physician. Using these rules (called protocols) the computer can automatically remind the physician of situations which require remedial action. The overall objective of this work is to determine whether these automatically-generated reminders can improve physician compliance with pre-defined standards of care, and thereby improve the quality of care in terms of patient outcomes. A controlled clinical trial will be performed to decide these questions using a randomized group of study and control physicians. The compliance rate will be measured by the rate of response of physicians to the occurrence of clinical conditions requiring certain actions. The patient outcome will be measured by a combination of measures, including physiologic (hematocrit, diastolic blood pressures, wellbeing, etc.), resource utilization by diagnosis, and patient costs.